On Fri, 25 Mar 2022, Ben Avison wrote:
Note that the benchmarking results for these functions are highly dependent
upon the input data. Therefore, each function is benchmarked twice,
corresponding to the best and worst case complexity of the reference C
implementation. The performance of a real stream decode will fall somewhere
between these two extremes.
Signed-off-by: Ben Avison <[email protected]>
---
tests/checkasm/Makefile | 1 +
tests/checkasm/checkasm.c | 3 ++
tests/checkasm/checkasm.h | 1 +
tests/checkasm/vc1dsp.c | 94 +++++++++++++++++++++++++++++++++++++++
tests/fate/checkasm.mak | 1 +
5 files changed, 100 insertions(+)
create mode 100644 tests/checkasm/vc1dsp.c
Actually, this test already paid off - thanks! It caught a real issue with
the existing x86 loopfilter assembly. The stride parameter is 'int', but
the assembly uses it as a full register without clearing/sign extending
the upper half.
Instead of complicating the assembly, the usual remedy is to change the
parameter to ptrdiff_t, to avoid the issue altogether - I'll send a patch
for that.
// Martin
_______________________________________________
ffmpeg-devel mailing list
[email protected]
https://ffmpeg.org/mailman/listinfo/ffmpeg-devel
To unsubscribe, visit link above, or email
[email protected] with subject "unsubscribe".
